Hi all,

First of all, thank you for reading this post. I do need some of your professional advice on my situation. I am a new PhD student in industrial engineering and I am so puzzled on the research direction I should pick.

The two available directions I am interested in now are 1) data mining/machine learning and 2) traditional OR.

On one hand, professor A focuses more on CS side and does a lot of machine learning application such as prediction and time-series analysis in healthcare. I believe this area is a lot hotter in the coming 10-20 years in the industry. On the other hand, professor B is experienced in traditional OR side. He is the expert in response surface modeling, statistics, and dynamic programming. He exactly knows what should be done for each project, and he knows well how to integrate and apply bunch of complicated theory and knowledge upon a specific problem.

As for the advisers themselves - Professor A is encouraging and perhaps a little bit pushy, demanding result from you all the time. Otherwise he will not spend time on you. Whereas for Professor B, he is very professional and experienced in leading and giving one research directions.

In this regard, I love Professor A's research topic and Professor B's work style. I do want to work for Professor B but I am just afraid that I cannot get a job after graduation easily as his research topics (response surface modeling, statistics, and dynamic programming) are not very hot indeed in the industry. Also the research difficulty for Professor B's topics is obviously a lot harder.

If you were to make a blunt decision of choosing one of these two advisers, which one would you choose? I realize I can have co-advising arrangement but this is something I don't prefer. I don't want to follow Professor A on the first year and then follow Professor B on the second year. It shows you are not loyal enough to someone.

Thanks, again, for reading this long post. I do look forward to your constructive comment!

Sincerely, Mr.F

asked 25 Aug '14, 00:01

bookyeah1679's gravatar image

accept rate: 0%


+1 I never liked these career questions, but I think this is the first well thought out and precise academic career question we have had on ORX.

(25 Aug '14, 02:27) Bo Jensen ♦

Are you planning to go into industry or an academic position after graduation?

(25 Aug '14, 15:11) Paul Rubin ♦♦

I wanna go to the industry instead of being a professor after graduation. So is there any advice you can share with me in this regard?

(25 Aug '14, 15:21) bookyeah1679

Disclaimer: the last time I interviewed for an industrial job (or any job, for that matter), punch cards were still the main input method for computing.

That said, I think that companies hiring PhDs in OR or related fields are more interested in your skill set, how well you interview and what your references have to say than in your dissertation area. Picking a dissertation topic based on what you think is hot is risky -- areas heat up (and cool off) fairly quickly. Some companies may be looking for people with specific skills in "big data analytics" (or simulation, or optimization, or ...), but many companies want to know whether you can look at a business problem and come up with a model/approach for it (which could be any of those subjects, or a combination of them), and then carry that idea through to a workable solution.

I think that, similar to what @semi said, I would view coursework through the eyes of a decathlete (be good with a number of tools, without becoming overly specialized in any of them), and for the dissertation pick something I think I will enjoy working on or someone I think I will be comfortable working for (whichever is more important to you personally), without too much concern about the connection between dissertation topic and job market.

If you decide to specialize in something "marketable" (say data mining), be sure you like it -- because if you do that for your dissertation, and then get hired based on that, you're going to be doing it for a while!


answered 26 Aug '14, 18:09

Paul%20Rubin's gravatar image

Paul Rubin ♦♦
accept rate: 19%

It seems that interest is more important than career progression. Thanks for your advice, Professor Rubin. Let's keep in touch!

(27 Aug '14, 01:57) bookyeah1679

As a person who did PhD with traditional OR and who did post doc with prediction model (machine learning) in healthcare, I agree with Paul, Absolutely. Also think about which is easier, "OR expert learns machine learning" vs. "Machine learning expert learns OR". Obviously, it depends.

but, i believe this is also important.

(28 Aug '14, 14:57) ksphil

so which one do u think is easier, "OR expert learns machine learning" and "Machine learning expert learns OR"?

(05 Sep '14, 20:14) bookyeah1679

@bookyeah1679, /re @ksphil: You also might be interested in reading this blog post, "Operations Research, Machine Learning, and Optimization" – by @tdhopper.

(06 Sep '14, 04:10) fbahr ♦

My 2cents:

The analytical skills you would acquire with state-of-the-art OR would allow you to pick up on any data mining technology when needed in your career; it is rather challenging to pick up OR theory once you start an industrial career. Additionally, I would recommend you aim at mastering both disciplines, by choosing OR topics which involve data mining tasks/challenges. There are plenty.

As a side advise, I would stress the importance of coding, and more specifically the ability to pick up on any new technology and not neglecting coding methodology.


answered 25 Aug '14, 09:58

semi's gravatar image

accept rate: 0%


+1 for coding skills. I am probably not getting too popular for saying, that I think it's harder to learn CS later on than OR. The problem is low CS skills is a complete show stopper, you don't produce anything, even you have great ideas, no one is going to implement them for you. When I finished school, my coding skills were best among my fellow students, little did it help me, since I was way behind my new co-workers. That was a tough uphill battle for years to catch up.

(25 Aug '14, 12:11) Bo Jensen ♦

I agree coding is important, but how important depends on industry v.academy and small (do your own coding) v. large (delegate it to the IT department or to a graduate student).

(25 Aug '14, 15:10) Paul Rubin ♦♦

If you want to go to industry you must understand that your employer will want quick return on investment. It means that you must have hands on experience on real problems.

Even though data analysis of all kind is a hot topic now I do believe that optimization has a bright future in industry. Therefore, my advice is to select a topic you like, as opposed to the hottest topic today. But make sure you learn how to code (program) and use popular tools during your PhD. I would second Bo's advice that CS course are definitely a plus.

This is from someone who hired 40+ PhDs in industry over the years.


answered 01 Sep '14, 14:43

jfpuget's gravatar image

accept rate: 8%


Although I'm totally not in the position to object (and I'll admit that the statement "your employer will want quick return on investment" is arguably true), IM_H_O, bringing "choice" to the table (i.e., being familiar w/ a variety of "tools") is more valuable than "mastering" one specific [popular] tool. (Might depend on the job description/requirements, though.)

(01 Sep '14, 16:14) fbahr ♦

I agree, and that's why I said tools (plural).

(02 Sep '14, 03:30) jfpuget

This may not be an option for you given that you already narrowed down your choices somewhat (which is a good thing to do), but the interface between Data Mining and Optimization is rich in research problems, and getting a lot of attention. Just take a look at the abstracts for the last KDD. In particular, solving large data mining problems using distributed optimization will get you both great papers and great industrial offers. On the riskier side, there is the wide open field of using Data Mining techniques to analyze the output of optimization procedures.


answered 07 Sep '14, 20:17

Leo's gravatar image

accept rate: 8%

edited 09 Sep '14, 00:54


You beat me to the pole, I wanted to add an answer saying exactly this!

(08 Sep '14, 06:17) jfpuget

@jfpuget, You were the first person I thought of as I wrote this.

(09 Sep '14, 00:55) Leo
Your answer
toggle preview

Follow this question

By Email:

Once you sign in you will be able to subscribe for any updates here



Answers and Comments

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "Title")
  • image?![alt text](/path/img.jpg "Title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported



Asked: 25 Aug '14, 00:01

Seen: 1,625 times

Last updated: 09 Sep '14, 00:55

OR-Exchange! Your site for questions, answers, and announcements about operations research.